Update Dockerfile

Dockerfile

@@ -1,38 +1,59 @@
-# Use a lightweight Python base image
+# -------------------------------
+# ✅ Base Image
+# -------------------------------
 FROM python:3.10-slim
 
-# Disable CUDA since we're using CPU-only
-ENV TORCH_DISABLE_CUDA=1
-ENV TRANSFORMERS_CACHE=/app/hf_cache
-ENV HF_HOME=/app/hf_cache
+# -------------------------------
+# ✅ Environment Setup
+# -------------------------------
+ENV PYTHONUNBUFFERED=1 \
+    TORCH_DISABLE_CUDA=1 \
+    TRANSFORMERS_CACHE=/app/hf_cache \
+    HF_HOME=/app/hf_cache
 
-# Set working directory
 WORKDIR /app
 
-# Copy dependency list
-COPY requirements.txt .
+# -------------------------------
+# ✅ System Dependencies
+# -------------------------------
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    git \
+ && rm -rf /var/lib/apt/lists/*
 
-# Install dependencies efficiently
-RUN apt-get update && apt-get install -y git && \
-    pip install --no-cache-dir -r requirements.txt && \
-    rm -rf /var/lib/apt/lists/*
+# -------------------------------
+# ✅ Install Python Dependencies
+# -------------------------------
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
 
-# Copy application code
+# -------------------------------
+# ✅ Copy Application Files
+# -------------------------------
 COPY . .
 
-# Make cache folder writable
+# -------------------------------
+# ✅ Prepare Cache & Permissions
+# -------------------------------
 RUN mkdir -p /app/hf_cache && chmod -R 777 /app/hf_cache
 
-# Expose port
+# -------------------------------
+# ✅ Expose Port
+# -------------------------------
 EXPOSE 8080
 
-# Preload model embeddings (optional)
-# Commented out to speed up Cloud Build — will load at runtime in main.py
-# RUN python -c "import torch, pandas as pd; from sentence_transformers import SentenceTransformer; \
+# -------------------------------
+# ✅ Health-friendly Startup
+# -------------------------------
+# NOTE:
+# Model & embeddings are loaded lazily in a background thread (main.py)
+# so container becomes healthy immediately and avoids 30-min timeout.
+#
+# You can preload if you really want, but it’s slower on build:
+# RUN python -c "from sentence_transformers import SentenceTransformer; \
 # model = SentenceTransformer('./muril_combined_multilingual_model'); \
-# df = pd.read_csv('./muril_multilingual_dataset.csv').dropna(subset=['question','answer']); \
-# embeddings = model.encode(df['answer'].tolist(), convert_to_tensor=True); \
-# torch.save(embeddings, './answer_embeddings.pt'); print('✅ Precomputed embeddings saved');"
+# print('✅ Model preloaded')"
 
-# Run the FastAPI app with uvicorn
+# -------------------------------
+# ✅ Start FastAPI with Uvicorn
+# -------------------------------
 CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8080"]